Research Summary & Projects

Introduction

Our interests involve developing and utilizing coarse-grained and efficient computational methods to simulate biomolecular nanomachines. The network methods employed include the (A) elastic network models (GNM & ANM) and (B) constraint network model of FIRST. What follows are several brief descriptions of several different applications for these methods. This includes Protein Folding Cores & Pathways, Biomolecular Dynamics, and Multi-scale Modeling.

Protein Folding Cores & Pathways

My research using the above network models of proteins has suggested that the network of intra-protein contacts contains information critical for reproducible folding. Comparison with experimental data including H-D exchange data, mutations, and Φ-value analysis confirms this hypothesis. The next challenge is use this information in conjunction with other data to predict and refine three-dimensional structures of proteins.


Biomolecular Dynamics

Even when a static structure is well known, a more complete understanding of biological function requires elucidating the structure’s likely biomolecular dynamics. In cases where structural knowledge of a given protein is incomplete, application of structural preferences deduced from the protein folding studies mentioned above will provide better structural models that can then be used as a starting point for exploring potential motions. Although biological functions can often be inferred for one molecule from a homologous one, the precise steps describing such processes require more detailed analysis.

Using the elastic network method sketch above, a few global modes are able to characterize biologically relevant motions. For example, in ribosome, the ratcheting associated with translation is captured and in the HK97 bacteriophage virus capsid, maturation from prohead to mature form is observed.


Multi-scale Modeling

Presently there is a large disparity between what can be measured experimentally and simulated computationally. Because biological processes occur on a wide range of time and length scales, no single method is able to capture all motions and functions. These fluctuations can involve ampltitudes smaller than an Angstrom on the order of femtosectionds or move several millimeters over the span of hours. Thus bridging the gap between various experiments and simulation methods for these time scales requires the development of a new computational framework which embeds the fine atomic-level detail into a hierarchy of various coarse-grained networks. The dynamics for a particular type of motion can then be deduced by analysis of the corresponding network.